Critical role of butyrophilin 3A1 in presenting prenyl pyrophosphate antigens to human γδ T cells

T lymphocytes recognize antigen via the heterodimeric T-cell receptor (TCR) which is closely associated with the signal-transducing CD3 complex. αβ TCR-expressing T cells recognize peptides presented by MHC class II (for CD4⁺ T cells) or MHC class I molecules (for CD8⁺ T cells). In striking contrast, the majority of human γδ T cells expressing the Vγ9Vδ2-encoded TCR recognize small non-peptidic phosphorylated ligands which have been identified as pyrophosphate intermediates of microbial and eukaryotic isoprenoid synthesis. However, it has been unknown for many years how such prenyl pyrophosphates are presented to the γδ TCR. In a recent paper published in Nature Immunology, Vavassori and co-workers have identified butyrophilin 3A1 as the elusive cell surface molecule that can directly bind pyrophosphate antigens for presentation to the γδ TCR.¹

γδ T cells are a minor subset of T lymphocytes in the peripheral blood, but occur at higher frequency in mucosal tissue. Based on the expressed TCR repertoire, human γδ T cells come in two flavors, i.e., γδ T cells expressing Vδ2 (almost exclusively paired with Vγ9) and non-Vδ2 T cells (which usually express Vδ1, frequently paired with Vγchains other than Vγ9). In healthy adult individuals, the majority (anywhere between 50% and 95%) of peripheral blood γδ T cells express a Vγ9Vδ2 TCR.² Vγ9Vδ2 T cells display potent cytotoxic activity, produce a range of cytokines and are thought to play important roles in microbial and tumor immunity.² The ligands that are recognized by the Vγ9Vδ2 TCR have been identified as low molecular weight prenyl pyrophosphate intermediates in isoprenoid synthesis. Microbes use the 2-C-methyl-𝒹-erythriol 4-phosphate pathway to generate (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP) which selectively activates Vγ9Vδ2 T cells at pico- to nanomolar concentrations. Eukaryotic cells use the mevalonate pathway of isoprenoid synthesis, in which isopentenyl pyrophosphate (IPP) is an intermediate that also activates the very same Vγ9Vδ2 T cells, even though 3-log higher (i.e., micromolar) concentrations are required.³ Stressed and transformed cells frequently have a dysregulated mevalonate pathway and produce higher amounts of IPP which marks them for recognition (and killing) by Vγ9Vδ2 T cells.⁴ Interestingly, the endogenous levels of pyrophosphate production can be manipulated by clinically used drugs, notably aminobisphosphonates (n-BPs) and statins. n-BPs are used to treat osteoporosis and bone metastasis in some types of cancer. n-BPs increase the intracellular accumulation of IPP by inhibiting the IPP-processing enzymefarnesyl pyrophosphate synthase. As a consequence, the treatment of tumor cells with n-BP augments their susceptibility to γδ T cell-mediated killing.⁴ In contrast, statins (which are used to lower cholesterol levels and to prevent cardiovascular disease) inhibit HMG-CoA reductase upstream of IPP synthesis and therefore, reduce intracellular IPP levels leading to diminished sensing by Vγ9Vδ2 T cells.

The recognition of both microbe- and host-derived prenyl pyrophosphates (here collectively termed ‘phosphoantigens') provides a basis for the dual role of Vγ9Vδ2 T cells in microbial and tumor immunity. While it was convincingly shown (e.g., by TCR gene transfer) that the recognition is TCR-dependent,³ it remained unresolved for many years how such phosphoantigens are presented to the Vγ9Vδ2 TCR. It became clear, however, that presentation (if at all required) did not involve conventional HLA class I/class II or CD1 molecules. A first hint for the existence of species-restricted presenting molecule(s) was provided with the observation that mouse cells could not substitute for human cells in the presentation of phosphoantigens to human Vγ9Vδ2 T cells.⁵ Interestingly enough (and in line with the absence of presenting cells in this species), there is no homologous TCR in the mouse that can recognize prenyl pyrophosphates. Recently, substantial progress has been made in the field, and a series of recent publications has identified a crucial role of a specific member of the butyrophilin family in the activation of Vγ9Vδ2 T cells by phosphoantigens.

Butyrophilins (BTNs) are type 1 transmembrane proteins with two Ig-like domains (IgC, IgV) in their extracellular part that are structurally related to B7 costimulatory molecules. Some BTN members carry an intracellular B30.2 domain (Figure 1a). In humans, the BTN genes are clustered in the MHC class I region on chromosome 6. The subfamily BTN3 has been assigned a CD number (CD277), but in fact, it contains three closely-related genes, BTN3A1, BTN3A2 and BTN3A3.⁶ Monoclonal anti-CD277 antibodies were evaluated for modulatory effects on T-cell activation, with seemingly contrasting results. Using monoclonal antibody (mAb) 232-5 raised against BTN3A3, Yamashiro et al.⁷ reported inhibition of T-cell activation when this mAb was added in soluble form to the cell cultures.⁷ In contrast, Messal et al.⁸ observed costimulatory effects when a different anti-CD277 (clone 20.1) was immobilized together with anti-CD3 mAb to stimulate human T-cell proliferation.⁸ Using this very same anti-CD277 mAb, Harly et al.⁹ recently reported a selective expansion of Vγ9Vδ2 T cells when peripheral blood mononuclear cells were cultured in the presence of mAb 20.1 and IL-2. In contrast, a different anti-CD277 mAb (clone 103.2) completely inhibited the Vγ9Vδ2 T-cell activation in response to tumor cells or n-BP-treated cells.⁹ By shRNA-mediated knockdown, they could further show that activation of Vγ9Vδ2 T cells was critically dependent on the BTN3A1 and not other isoforms of BTN3. In contrast to BTN3A2, BTN3A1 carries an intracellular B30.2 domain which was shown by Harly et al.⁹ to be essential for n-BP-mediated γδ T-cell activation. Together with further experiments, the authors concluded that phosphoantigens induce changes in the mobility and conformation of BTN3A1 by interfering with the intracellular B30.2 domain. Such alterations would then be sensed (in a not precisely defined way) by the Vγ9Vδ2 TCR. These researchers further argued that agonistic mAb 20.1 mimicks the induction of such alterations of the BTN3A1 molecule (Figure 1b); they did not obtain evidence for direct binding of phosphoantigens to BTN3A1.^9,10 Very similar conclusions were reached in another recent study by Wang et al.¹¹ Like Harly et al., Wang et al. also noticed the selective activation and expansion of Vγ9Vδ2 T cells by mAb 20.1. They confirmed the essential role of BTN3A1, but not BTN3A2 and BTN3A3 for γδ T-cell activation by IPP. Extensive structural studies performed by these authors did not reveal a binding site for pyrophosphates in the extracellular domain of BTN3A1 but predicted a basic pocket in the intracellular B30.2 domain. However, using a photoaffinity pyrophosphate labeling assay with recombinant BTN3A1 protein the authors could not detect significant high affinity binding, arguing in their view against a direct phosphoantigen-presenting function of BTN3A1.¹¹

Figure 1.

Role of BTN3A1 in phosphoantigen presentation. (a) BTN3A1 consists of two Ig-like extracellular domains (IgV, IgC), a transmembrane region and an intracellular B30.2 domain. (b) Multimerization of BTN3A1 by agonistic anti-CD277 mAb 20.1 decreases lateral motility and induces a 19-Å rotational shift^9,10 which might somehow contribute to γδ T-cell activation. (c) As shown by Vavassori et al.,¹ the extracellular IgV domain of BTN3A1 can directly bind HMBPP and IPP, associated with conformational changes and direct interaction with the Vγ9Vδ2 TCR. The role of the intracellular B30.2 domain (if any) in this setting has not been determined. (d) It remains presently unclear how endogenously generated phosphoantigens (e.g., IPP accumulating in response to n-BP treatment) are ‘presented' by BTN3A1. Intracellular IPP might directly bind to B30.2 and thereby, change the conformation of extracellular IgV domain.⁹ Alternatively, intracellular IPP might be secreted and then bind directly to BTN3A1 on the same or neighboring cells. HMBPP, (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate; IPP, isopentenyl pyrophosphate; mAb, monoclonal antibody;n-BP, aminobisphosphonate; TCR, T-cell receptor. ▴, HMBPP or IPP phosphoantigens; , anti-CD277 mAb 20.1.

A major breakthrough in the field was recently published in Nature Immunology.¹ Searching for the elusive phosphoantigen-presenting molecule, de Libero's group followed a different approach. Arguing that human but not mouse cells can present IPP and other phosphoantigens to Vγ9Vδ2 T cells, they analyzed somatic mouse–human cell hybrids which had retained selected human chromosomes. Whenever such hybrids had retained human chromosome 6p, they could activate mouse lymphocytes expressing a transgenic human Vγ9Vδ2 TCR in the presence of HMBPP.¹ Mouse TCR transgenic cells were used as a readout system in order to avoid possible interference of self-presentation when using human Vγ9Vδ2 T cells. Further step-by-step analysis including comparative gene arrays of phosphoantigen-presenting cells narrowed the number of candidate genes to below 20, of which some encoded butyrophilin family members. By using siRNA-mediated downregulation and retransfection of various BTN3 isoforms, Vavassori et al. then also identified BTN3A1 as the critical butyrophilin family member for activation of phosphoantigen-reactive γδ T cells.¹ In addition, however, they were able to demonstrate direct binding of HMBPP and IPP to the recombinant immunoglobulin V-like domain of BTN3A1 (Figure 1c) and furthermore, they crystallized the complex of BTN3A1 IgV-like domain with the bound IPP and HMBPP at a resolution of 2.0 and 1.9 Å, respectively. These studies identified a shallow groove in the distal domain of BTN3A1 for low affinity binding of pyrophosphates and a critical role of specific residues such as Lys39. Finally, these authors could demonstrate binding of soluble multimeric Vγ9Vδ2 TCR to immobilized BTN3A1 which was further increased by IPP. Taken together, these studies have identified BTN3A1 as the long sought after molecule presenting phosphoantigens to the human Vγ9Vδ2 TCR.

Has the mystery of phosphoantigen presentation to human Vγ9Vδ2 T cells now been fully solved? Well, the results of Vavassori et al.¹ have convincingly demonstrated the direct binding and presenting function of BTN3A1, in contrast to the two other studies which had implicated an indispensable but indirect function of BTN3A1.^9,10,11 In addition to the structural results, the fact that fixed BTN3A1-expressing cells can present phosphoantigen to Vγ9Vδ2 TCR transgenic murine cells, unambiguously demonstrates the direct phosphoantigen presenting capacity of BTN3A1, independently of conformational alterations possibly initiated through the intracellular B30.2 domain.¹ However, several issues remain unclear at this point and are open to further investigation. For instance, it is not clear how intracellularly generated IPP (e.g., following n-BP treatment) can be presented by BTN3A1.Wang and co-workers¹¹ identified a basic pocket in the B30.2 domain that could perhaps bind prenyl pyrophosphates with low affinity, possibly leading to conformational changes of the extracellular BTN3A1 domains. Alternatively, IPP might be secreted, and it could then directly bind to the extracellular IgV domain of BTN3A1 on the same or neighbor cells for presentation to Vγ9Vδ2 T cells. These various scenarios are illustrated in Figure 1d. Moreover, the precise function of the intracellular B30.2 domain remains unknown. In addition to a possible role in intracellular phosphoantigen sensing, the B30.2 domain can mediate protein–protein interactions¹² and might be required for trafficking of BTN3A1 to distinct intracellular compartments.¹ Even though some open questions remain, the recent identification of the central role of BTN3A1 in phosphoantigen presentation has tremendously advanced our knowledge in this area, a mystery ever since the first discovery of microbe-derived phosphoantigens for the human Vγ9Vδ2 T cells.¹³